Keyword Research Guide: How to Find Terms Worth Writing For
Most keyword research goes wrong before anyone opens a tool. People type their product name into a keyword planner, sort by volume, and pick the biggest number. Then they wonder why a six-month-old site selling handmade candles never ranks for "candles".
Good keyword research is closer to customer research. You are trying to find the exact words people type when they have a problem you can solve, and then work out which of those searches you have a realistic chance of winning.
Start with what customers actually say
Before any tool, write down 15 to 20 phrases from real conversations. Look at WhatsApp enquiries, support emails, questions people ask in your shop, comments on your Instagram posts and reviews of competitors.
A bakery in Jaipur might hear "eggless cake delivery tonight", "fondant cake price for 2 kg" and "birthday cake near Malviya Nagar". None of those look like the generic term "cakes", and all of them are closer to a sale.
These are your seed terms. Tools are good at expanding seeds. They are bad at telling you which seeds matter to your business.
Expand the list with tools and Google itself
Put your seeds into whatever keyword tool you use. Google Keyword Planner is free with a Google Ads account, and paid tools like Ahrefs, Semrush and others give more detail. Then use Google directly:
- Autocomplete. Type a seed and note the suggestions. Add a letter after the seed ("eggless cake a", "eggless cake b") to see more.
- People Also Ask. Click a few questions and more appear. These are ready-made subheadings and FAQ entries.
- Related searches at the bottom of the results page.
- Google Search Console, if your site already has some traffic. The Queries report shows terms you already appear for, often ones you never targeted.
Dump everything into a spreadsheet. Don't filter yet.
Group keywords by intent, not by wording
Next, sort the list by what the searcher wants. Four buckets cover most cases:
| Intent | Example | What the page should be |
|---|---|---|
| Informational | how to store fondant cake | Guide or blog post |
| Commercial | best cake shops in Jaipur | Comparison or list |
| Transactional | order eggless cake online | Product or category page |
| Navigational | [brand name] menu | Your own branded page |
Keywords with the same intent and nearly the same meaning belong on one page. "Eggless birthday cake" and "birthday cake without egg" are one page, not two. Getting this grouping right early saves you from keyword cannibalization later.
Read the search results before you commit
This is the step people skip, and it is the most useful one. Search each main keyword and look at the first page:
- What type of pages rank? If all ten are product pages, your blog post will struggle.
- Who ranks? Big marketplaces and national brands, or small local sites and forums?
- How old and how thorough are the ranking pages? Thin, outdated pages are an opening.
- What extra features show up? Map packs, videos, shopping results and AI-generated summaries all push organic results down.
If Reddit threads and small blogs sit in the top five, a well-written page from a newer site can compete. If every result is a household-name brand with thousands of referring domains, park that keyword for later.
Judge difficulty with your own eyes
Tools give each keyword a difficulty score. Treat it as a hint, since each tool calculates it differently and mostly from backlink data. Our guide to keyword difficulty explains how to sanity-check those numbers. The short version: compare the ranking pages to what you can realistically produce and promote in the next few months.
Volume matters less than you think
A keyword with 50 searches a month that brings buyers beats one with 5,000 that brings browsers. For a local bakery, "cake delivery Vaishali Nagar" at low volume is worth more than "chocolate cake recipe" at high volume, because recipe searchers want to bake, not buy.
Search volume figures are also estimates and are often grouped into ranges, so don't treat 90 versus 110 as a real difference.
Turn the list into a plan
You should now have groups of keywords, each with an intent and a rough difficulty. Pick:
- Quick wins: low difficulty, clear business value. Write these first.
- Core pages: your main product or service terms. These need the best content and, usually, links.
- Long-term targets: competitive terms you will build toward through topical authority and links.
For each page, record the main keyword, two or three close variants, the intent, and the URL it will live at. That single sheet becomes your content calendar.
Common mistakes
- Choosing keywords only by volume.
- Writing a blog post for a keyword where Google clearly wants a product page.
- Creating separate pages for tiny wording differences.
- Ignoring Search Console data once the site has traffic.
- Never revisiting the list. Re-run research every few months, because new questions appear and old ones fade.
Putting it into practice
Open a blank spreadsheet, write down 15 phrases your customers have used in the past month, and run each through autocomplete and People Also Ask. By the end of an hour you will have a better starting list than most sites get from a week of staring at tool exports.